NVIDIA’s Agentic AI Push | Coding in the Age of AI | Big Tech vs AI Scams

Boom. Hey everyone, happy Tuesday. You know, the days are long, the hours are short.

I, I gotta tell you, I wake up at about 6:00, 6:30 in the morning, and the first thing I do is check my email for what my AI has sent me over the night to get me ready for the day, and I start, like, looking at today's news and everything. And before you know it, man, my dog is barking to go out. I gotta get her out, I gotta get dressed, I gotta get to the office.

There's just not enough hours. I thought this AI's supposed to make our lives easier and do work for us. But, it's exciting.

It's exciting. We've got an exciting gang to talk about these things with you today. Let me introduce you to who we've got.

Mr. Scarf himself, Chris Blask. The one and only, Nate Scarsella.

Our... I didn't realize we had a bureau chief, but he's our bureau chief. Mike, Mike gave me that title.

Oh, okay. Well, it's whatever. Well, he's our- It's your remote and only ...

he's our West Coast bureau chief, which means we must have an East Coast bureau too. Right. But the one and only, Jon Swartz, and of course my friend, Mitch Ashley.

Hold... I thought you were at the, the, GTC thing, Mitch. No, nope, John already had it covered, and they called and said, "Yeah, we got John.

" Exactly. Well, once they have the bureau chief, it's a big step down from there. Or you.

We got the bureau chief. Oh, you got the bureau chief. Wow.

Wow. It sounds like something out of J. Edgar Hoover or something.

I don't know. That's smart, yeah. But, anyway, gang, welcome.

Welcome today. We've got a lot to go over. Of course, the big news is seems, does, is seem to be coming out of this NVIDIA GTC event.

I heard it was, like, a two-hour plus keynote or something. Jon, why don't you kick it off? Okay.

Yeah, no, i- it's funny. This... You know what this, that event yesterday reminded me of?

And it does remind me of, of the event they did at CES. Back in the day, and still to this day, i- the second Tuesday in September is when Apple does its big event for, the hardware stuff, and that dominates the news cycle. No one else announces that day, and that's pretty much what happened yesterday at GTC.

So I mean, there is flooding the zone, and then there is submerging it, and that's what NVIDIA did. They held an event at the SAP Center, where the San José Sharks of the NHL play. I kid you not, it was packed to the rafters.

There were about at least 15,000, probably 17,000 people there for a two-hour address by Jensen. Uh, it, it... Just to kind of to step back, on Sunday I did a pre-briefing with NVIDIA among a bunch of other reporters.

We were flooded, Mike and I were flooded with, related GTC news. So no one else is dare make an announcement that day, and there's a reason why, is NVIDIA is the it company. It continues to be.

And, one other thing I'm gonna mention before we go on, it's, it's something that I think is really interesting, is I sat at an angle where I could see the teleprompters, 'cause I always w- I'm interested in how these guys handle this stuff. With Jensen Huang, what he does, they show a chart, and he speaks. There is no, there's no script.

He did this for two hours, and it was flawless. It was amazing. Um, so and any event, he s- he kicks off what he calls the Super Bowl of AI, GTC, and there are two things that jump out at me, because there are so many.

I'm gonna leave it up to you experts to kinda dive into what interests you. But one of the things that really got my attention was the Vera Rubin platform, was this project that they're using to transition from simple generative models to agentic AI. Um, he also announced, Jensen, the birth of this ne- quote, "new renaissance in software," which is a suite of open source tools that would shift AI from a passive chatbot to an autonomous workforce.

It's called the NVIDIA, and I have to read this, Agent Toolkit. There were a constellation of companies up there, Adobe, Salesforce, Microsoft, ServiceNow, Dell, Micr- Microsoft. You get the, you get the, the, you get the, the gist of it.

They were all up there talking about how they're going to standardize agentic AI, especially for enterprise. Um, one other thing that Jensen mentioned which got a lot of headlines, and I think it is really significant, is they're projecting a trillion dollars in sales through Blackwell and Rubin in 2027, which is twice what they had estimated in October. " Um, I know Mitch wrote about it.

I probably should hand it off to Mitch. I, we, I used a comment from Mitch, who was following this closely. But this was pretty impressive, but it also, it, it just shows you the power and the strength and the scope of NVIDIA.

You know, to the, to the point, this is, this conference has become what Steve Jobs and the Apple conferences were. Yeah. He was, you know, at his height of doing iPhone and other announcements, so it really is the event that, that, that appears that you don't cross until it's over.

Um, so another, interesting thing, first, first reaction was it reminded me a tiny bit of the AW- AWS, conference in December, where they started out talking about software, like you're a hardware company. A- and it was a long talk about software. And Jensen, Jensen, sort of like he got religion about not just software, but about apps.

Y- yes. Coming up the stack to saying, "Oh yeah, there's a reason why we're doing all this stuff. " You know, kinda, hey, it's kinda like it's the economy, stupid.

You know? It's the same, it's the same thing. Kinda thing like, ahaThere, there's a greater purpose here than just building data centers and hardware.

There's a reason that we're doing that. And he did almost describe the hardware layer as becoming the AI factory, the token factory, right? That's, that's what all this hardware about, is about, is, is creating token budgets for people to do things with, particularly software developers.

And so the number of announcements around that, this, Agent Toolkit is what caught my attention right away because I expected a great, you know, a great deal of talk about NeMo Claw, which he did talk about later quite a bit. Um, but he started out talking about things like in Agent, Agent Toolkit is a, an o- an open shell, which is an open source project they've created. I hope they contribute that to Linux Foundation or Agentic AI Found- Foundation, which is a containment environment to run this stuff in, right?

First question is, how do we control what it does? How do we observe it? How do we govern it?

All of those kind of things. Um, then there's an, an AIQ part of it, for doing some of that management, the sec- the s- governance kind of activities. Some of the, the elements of an agent control plane, and that's happening across the stack.

Up and down, left and right in the development cycle, people are issuing their ideas of what controlling agents or observing agents or, making agents accountable. So hi- his, his answer was, "We got it covered. We're, we, we got it all good.

It's all right here. " So not quite the case, but you know, some, some great stuff. So that, that was all, you know, like love language to me, right?

I'm like, "Good. Okay. " Um, he did go into great deal about, Neo Claw.

And as I was listening to that, this-- every year, kind of beginning of the year, it seems like we hit the what is the big thing that's gonna swing the market? You know, first it was, OpenAI or ChatGPT, then along came MCP last year, right? " Well, that's so what's happened at least in, in, Jensen's, and I think there's some truths to this, is this is the era where OpenClaw and things like it that let us do so much more quickly with agents, not always securely, not always governed, but that is the new springboard for creating all this stuff.

Um, and Alan and I have talked a lot about this stuff. I know he, he's sitting there very calmly, but there is a storm erupting, and I'm just gonna let-- unleash Alan and let him go. It's a quiet storm.

A quiet storm. But, no, Mitch, I think your last line was, was dead on there in that... Look, the real message here is agentic AI is here.

Mm-hmm. Generative AI was great, is great. It's the basis that we build the agentic stuff on.

But people are no longer going let the word go forth, right? People are no longer gonna be satisfied just asking your generative AI a question and having it spit out an answer or a document for you. People want their AI to do work for them, real work, and they are using agents.

And this OpenClaw thing, OpenClaw is to agents what ChatGPT was to generative AI, what Netscape was to web browsing, right? That's the best analogy I could give you. And I think that cat's out of the bag and not even Jensen can hold on to that tail.

Excuse me. The real battle is who controls this stack? Who controls this cat and who controls this stack?

The cat's open source, but like all open source, you're gonna get some people who enjoy doing the open source thing and kind of building it themselves and tinkering and tweaking and... Like my friend Chris, Mitch, you guys are tinkerers and tweakers, tweakers. Then there are gonna be people who want something a little more polished, you know, based on open source certainly, but a little more co- polished, and we're gonna-- and I want it to just do things for me, right?

I mentioned yesterday, I spent the weekend on Perplexity Compute and rearranged a good chunk of Textstrong's workflow and how we're doing it. I didn't have to set up the Mac Mini. I didn't have to install a bunch of things.

It, it kinda... You just tell it to install it and it does. I think Jensen realizes he wants both...

'Cause look, when your company's five trillion dollars, why not? He wants both parts of that puzzle. He wants to give the tinkerers the platform to tinker on.

He wants to give the people who want that polished kind of solution ready to do work, he wants to give them that too. He wants it secure, right? There was a big chunk of the JFrog people securing it.

I, or helping to secure the Nano Claw, whatever it's called. Um, I just interviewed the CEO of the company Open, OpenHand, OpenHand Index, who is working pretty closely with them on this for the tinkerer part of things. People wanna build their own agent even.

Um, but this is, this is-- to me, that's really the story here, right? That it's all about agents and agentic AI now, and how we do them at scale, how we do them securely, how do we build governance in? " That's not gonna last long.

But it'll come. It'll come. And then I think, John, what you said, and Mitchell echoed, which is, look, no one wanted a...

It's not that they wanted to steal his limelight. They couldn't steal his limelight. Whatever you put out while he's doing it is gonna get drowned out.

But the follow-ons today and everything elseWe've seen a lot, you know, small companies like Dell, I mentioned JFrog as the other one- ... you know, trying to ride these coattails. A- and I think a lot of what they're announcing is, is also important.

It all goes to this agentic stack that we're developing real quick. But, you know, I don't know if Nvidia winds up owning the stack. I forget, not Nanotech, what's the name of the- Uh, NeMo claw?

NeMo. NeMo. NeMo claw, yeah.

Yeah, the whole NeMo stack, if you will, though. Mm-hmm. I don't think they will, no.

Is that developing? I don't. Who...

You know. You know, Mitch, Mitch, Mitch, Mitch mentioned something that was really interesting and very perceptive, is that, he is this, uh... Jensen almost seemed enlightened by this concept of apps.

He kept coming back to it during his long introduction in, about software, like i- the bell rung, and, I, I just found that... And then I was taking notes on what he was saying that interested me, and that always jumped, that really jumped out. But a trillion dollars, that jumps out at me.

When you got a trillion dollars in revenue- Oh, yeah. And that got all the business press coverage, yeah. You know, you got a, a trillion dollars in revenue, that takes this from maybe 5 trillion to 12 to 20 trillion dollars in valuation.

I'm not telling anyone to buy stock, but- Yeah ... just saying It does take a lot of Gen Elle room. You gotta kind of catch your breath there.

Yeah. Yeah. But let, let you know, I, I think we've covered a lot of the Nvidia piece of it, but what about some of the other announcements around it?

Anyone? Kate? Yeah.

So, just coming from the network background, I, I like Akamai. You know, that I think that, you know, overall, they have real insight. And I just want you to realize, I've been talking about, on the back shows here, about distributed, right?

I mean, you guys... Right? Mm-hmm.

And so I think we really see a distributed model that is just rising, from the ashes here. And Akamai has, that they know inference, doesn't scale well. And so, you know, they're very, very powerful from, on the edge, and I just have always liked them.

So they can deliver, I was looking at some of this, 60% lower latency and major cost reduction, when we, versus, centralized cloud inference. So, yeah. So I, I'm like Akamai for the win here.

A- and I mean, you know- So let me put this down. Kate, Kate wants Akamai. What can we put you down for there, Kate?

5,000 shares? 10... I feel like Jordan Belfort.

The Wolf of Wall Street. Um, but no, I, I, I agree with you, I think. So here, here's an interesting thing on that, though.

Is the edge gonna be an Akamai, Cloudflare, let's call it traditional edge, where it's just sort of cloud service, but a little closer to you? Or is the edge at my desk running a Mac Mini? Yeah.

That's what I... That's, that's where it's going. I really believe it very strongly, and I think it just makes a lot of sense, and you see people, you know, even building like that in, you know, as we go forward.

I mean, you know, take a look here at my, at my friend Chris, you know. We see more and more things will be on-prem, and, and I think it's really, really a bad choice to be building these huge data centers, as I've also talked about, because the distributed model at the end of the day is very, very strong, and we have seen that, you know? I, I mean, we have seen edge, basically super computers, like with the Tor network and things like that.

I, I, I think everything is gonna be going back, you know, to that type of modeling. Very cool. I mean, look, again, Jensen covers all his bets, right?

Right. They have, they have their little, you know, Mac Mini variant of- Yeah ... of their hardware and, and stuff as well, right?

It's brilliant. Yeah, it is. Little Mac Mini.

A little? That holds- Well- ... a trillion parameter model for 3, 320.

Well, I meant footprint wise. Oh, okay. Yeah.

Okay. Yes. E- exactly.

And, you know, so. Yep, but, but Dell made- The only thing I- ... announcements too, right?

Yeah. The only thing I have to add to this block, though, is, is, is it's interesting that I was gonna mention Akamai as well and, and you did, Kate. Right.

You know, as I look at this block, yeah, yeah, sure, yeah, like you say, on, you know, Jensen and have, you know, the, they, they're big kids, they can play their game, but I look at this and I see Akamai, and I agree with Kate- Yeah ... 1,000%. I, I think- There you go ...

I think everyone's gonna want a piece of this. I think the security guys are gonna fall over themselves. We're gonna see it next week at RSA, talking about securing and governance for agentic AI stacks and systems- Yeah ...

and Edge and everything else. Look, this is... I mean, you know what?

Instead of the black turtleneck, we'll be wearing leather jackets, but- Yes. Good ... this, this is what we, this is what we got to look forward to.

But we got to hop... I mean, this is the big story. Yeah.

You know, I want to announce, we, we put out a new newsletter at Techstrong today called the Techstrong Brief, and it's gonna come out a couple times a week, two, three times a week, I believe, but we might go to daily. And, and, it's gonna have the big story of the day, editor's notes, and then coverage across the entire Techstrong universe, if you will. And, I mean, it was, looking at the draft of...

You know, we're still putting it through its paces. Looking at the draft of today, it was just amazing how much news did come out of GTC. Yeah.

There's so much more we could talk about. Yeah. Cloud strike.

Yeah. Everybody. It could be the whole show, right?

It could easily. You could d- we could do a GTC recap and, and do three shows. Yeah.

Uh, but let's move on to our next topic, right? So this one here is f- well, it's an AI topic too, programming in the age of AI. Guys, look, it's-It's changed what being a programmer is, what being a coder is.

We're builders. Every... But here's the thing, we're all builders, right?

I don't know if being a, a coder or a programmer is such a thing anymore as being a builder. Uh, there was a good article in The Times that I riffed off of it. But Mitch, what do we, what do we got on this one?

You know, I think w- we've gone from talking about what the changes are to developers to actually living it and experiencing it. And I think that's... Certainly for me, I've talked about this for a long time about developers moving to the engineers of AI agent dri- driven development.

What does that mean? It means you're not spending any time or very little time in front of code. You're spending it in front of agents directing their work, a- addressing things that come up, but also stepping back and saying, "What are we building?

Why are we building it? " If you're a developer, you're talking more about the architecture that you want this done and the security requirements. If you're an end user, you're, you're maybe asking OpenCloud, whatever platform about security, what do we need to do?

Are you doing that for me? How am I doing that? So for builders, it just kind of depends on where you are in the spectrum.

I think, I think the... So that's, that's a massive change. And, and but actually you can do so much more right now with what's happening.

The amount of code that I've generated or co-created with, with AI is scary. Um, the, the big thing that I think folks are starting to realize, and I've also been talking about this, is the bubble, AI bubble isn't the market bubble, it's the bu- bubble of debt we create where we expect humans to review and approve what's done by AI, whether it's code changes, security of things, incidents, observability, information. Uh, there are not enough people on the planet, I've said this multiple times, in the w- on the planet, in the universe, whatever other alien life forms might exist, to review everything that AI is doing today and will do for us.

It's just not a practical matter. So the industry's recognizing that we can't just observe, we can't just report, we can't just tell you about things. We've got to start figuring out how we as, as in our niche of the industry, has to start solving problems, not just reporting them.

Not doing a better job of synthesizing, fixing them. Um, so it, we have to move to a report it fixed, report it-fixed model, which of course for all of us raises the, just like, you know, AI's gonna generate all my code for me. Well, wait, wait a minute.

Uh, AI's gonna secure this stuff for me? Wait, wait, wait a minute. We've got this trust h**p to get over.

Um, so it, it's really... We, we are in the thick of the storm of now realizing what this is gonna mean, not only for developers, but DevOps, platform engineers, security engineers. I think all of us farther down the, the supply chain or the, the life cycle change is now kind of starting to feel those pressures.

Uh, agreed. Agreed. Um, but y- you know, I think for developers it's acute right now.

Mm-hmm. They're living it, yeah. When we look at platform engineers, security professionals, guys, this is, this is a wave, a tsunami that is working its way through here.

So, you know, and, and by the way, the, this first topic came, you know... We started talking about it, Harness did the survey in terms of DevOps workflows. But as we look at, you know, what else is coming down the pike, it- it's, it's, it's fundamentally changing, I mean, the development life cycle, the s- the...

Tech is the tip of the spear of these changes, right? And we're... I have a report, an article I'm putting out on the, a bunch of research done on AI jobs and, and you know, how it's affecting jobs.

But probably, you know, tech is out ahead of, of how this is affecting. Chris, you want to say something? I, I was just gonna say, you know, everybody's always saying, you know, as vendors and, and providers, you know, quite often, you know, myself, the, the, the world is changing and yada yada.

And you can tell when it actually does because it makes patterns, right? And you see patterns in headlines, patterns on this show over the last year, right? And what we're talking about right now, Mitch, exactly the way you said it, you know, th- this is a real dev person thinking about it really hard saying, "Oh yeah, that's right.

" And it's almost month to month, almost week to week if you pay enough attention to it. The conversation moves a little bit forward. And I can...

I'll make a prediction. Three months from now to six months from now, we're past this and saying, "Of course, you know, you're not gonna have human oversight in all these places. Where do you really need it?

" Yeah. And that takes trust in a couple layers that is not common right now, and not warranted right now because we're not doing all the things right, but we have to get there. You can see it from here.

It forces it. It'll force us to do it, right, Chris? Yep.

You know, one, one of the, one of the articles in this section, and, and by the way, in the notes we'll s- I think we include all the links, but one of the articles is something I wrote about Starkist and tuna's that taste good. Charlie the tuna. Charlie the tuna- Yeah ...

with good taste, is there was an interesting, an interesting study hap- or an interesting case of 6,000 lines of c****y code, not necessarily maliciousNot necessarily security riddled, but lower quality code, bad code. 6,000 lines introduced into a complete LLM of billions, trillions of lines even, right? Um, was enough to knock this thing off its keister, not...

Off its axis. Not just in terms of programming, but it started exhibiting all kinds of bad behavior. And I...

And there's a lesson there too. The lesson is we're running as fast as we can headlong into this new future. And it, it's not...

It's far from perfect yet. Yet, I say. Maybe, I don't know if it'll ever be.

But we... I... Someone...

I mean, who is the adult in the room who's not running after this thing as fast as they can or saying, "Wait, wait, wait. What about governance? What about governance?

What about- Adult supervision? Is that what you're asking for? Yeah.

Wait, who's the adult? Do we have any adults left in the room? Don't you think it's the people?

Us? All of us? If, if it's us- You know?

I'm afraid because If it's us- It's Kate. It's Kate ... we're in trouble.

We're in trouble. That's what I'm trying to say. We're in trouble.

Because the... Why is tech the tip of the spear? Because we're all too eager to go try this and do this.

Yeah. Mm-hmm. Just because we can- We're like a ton of poor Chris Blatts.

Maybe we should. Is that what you're saying? Yeah.

I mean, if we're the adults, we got trouble. You know, this is like one day I woke up and I realized I wasn't the kid no more. And- ...

though in my mind I still am. At that moment. But if we're the adults, we, you know, are...

Do we... Are we willing to put on the brakes for governance, let's say? Kate, Chris, I'll throw it out there.

I don't think it's necessarily brakes. I think it's, you know, we proceed, cautiously. And I think at the end of the day what I look at is, not necessarily resilience.

I'm more around stability, and I think, you know, we don't like... We don't even like unstable people, much less unstable systems. You know what I mean?

So, you know- Speak for yourself ... speak for yourself. Well, some of us vote for them, but that's a whole other story.

Wow. I've known you around a- Alan for a long time. You've always liked unstable.

Yeah. Well, yeah, it- But we could talk about... I'm sorry, Chris, go ahead.

I was just gonna say, it... The, the most important lesson I learned in my career, and as much as it pains me to say this, is that Alan and Mitch are right. Right?

You know, 'cause you, you go and- Wait, can you say that again? As a security person- Can you say that again? No, only once.

Only one. But as a security person, you know, in the community, we're trained, we train each other to, to say, "Oh my God, this company's so stupid. They're not doing the right things.

" But to be clear, your company exists. It, it starts every morning and it's got the lights on, it's paying people to do things. Whatever choices you've made to that point are correct, are correct from a security perspective, because you're not...

You know, if I'm not talking to you about a breach, then you've made all the right choices to this date, whether I thought they were right or not. Right? And to the point of being adults, yeah, this...

The world... And Kate, you're right, we, we like stable things. People do, companies do, individuals do.

As we run into things that are unstable, we'll tend to choose the stable ones. So and whether it's, Alan, to your example, individual LLMs, you know, we don't trust individual anything as adults because as adults we know that that thing we were really counting on is going to fall apart, sink, catch on fire, or whatever. That's what adults do.

It's not about being perfect at all. And we've got a lot of fragile systems and over-centralization. We'll figure it out.

We may crash some things, some economies and businesses- ... but on average, it'll work out like it always does. Yeah.

You know- Well- ... that's comforting. Exactly.

Well, when, when systems change- You know, we build on, on- I'm sorry, Kate. Sorry. Go ahead, Mitch, please.

Um, w- systems change when constraints change. And what cr- constraints have we had in this area? Well, it's how much...

How fast can people produce work? It's always about that. So now we've taken away the constraint, or we've collapsed the constraint of how much code we can generate.

So the problem isn't now how much code can we generate, it's how do we handle that much code downstream? So the, the bubble, you know, theory of constraints, right, that we've talked about Goldratt many times, Alan, the, the, the bottleneck moves. I think what's, what's different the...

this time is how fast it's moved and how quickly that creates a verification debt in, in the case of, you know, reviewing things, whether it's code, or someday it will be other things, or today, down, down the line. The good news is it feels really uncomfortable, but it creates opportunity. That's massive opportunity for entrepreneurs, for companies, product people- Yeah ...

security engineers. The difference this time, now we have tools. We can go address some of these issues ourselves.

I'm gonna build an agent that will go handle and, and some... and tell me which of the five things I need to look at are the one that I really need to care about, right? We don't have to depend on somebody else's judgment.

We can put our filtering if we choose to, or I can say, "Hey, Chris, amazing. Uh, Kate, would you sha- share with me that skill that does that? 'Cause that's awesome.

" Kate? Well, I just think that we sometimes... We often think the idea of, of things not working or things, you know, this whole risk.

I, as a security person and risk, I- I've always hated almost that, because I think sometimes we, we need to build on what is working and, and drop the components that are not working. And I think that's what AI and, and developing in AI helps us to do at a speed that's really important. Security has changed a lot, and we often used to think about, "Oh, I need so much information.

" Now it's like, yeah, forget about that. We need to just keep going. And, and that's where I think the security will actually be in the speed at the end of the day.

And, and that probably seems like, you know, this oxymoron, but I, I think that's what we'll see. It's a good example of when constraints change. We think of today of, well, how do we fix it?

How do we fix that code, or how do we fix- Yeah ... you know, your example, Alan, of here's all this really bad code, we are, we are rapidly approaching a point where I'd say, "I would never fix that. " It, it is- Yeah ...

it is much more effective. I don't wanna waste my time figuring out- Right ... if you fixed it right.

Just go rebuild it, you know? That's exactly... Or y- you know, take what's working.

It... We need to really change our mentality around here, because we don't have the time to fix things. It's because things are moving so quickly.

So what's working? And let's build on that. Fair enough.

Speaking of security, though, y- I, I gotta be honest with you. When I saw this next one come up in my daily today, I thought it was a joke, right? The anti- The AI Anti-Scam Alliance.

Sounds like something Maxwell Smart and the Chief cooked up, don't it? Maybe. That's what I was gonna...

You're having. Um, yeah, it's the old AI Anti-Scam Alliance. Yeah, uh-huh.

Uh-huh. Um, Kate, what... Or Chris, which, what's this one on?

So in 2003, the great David Berlind, read him on Substack, started... He got, cranky about spam, and the Jam Spam, effort, coalition was started. And I somehow got finagled into chairing this b****y thing, and we were looking at, at, you know, how do you g- you know, val- same sort of thing we're, we're working on today, right?

You know, how do I know when a sender is a valid marketer as opposed to a scammer and a spammer, or just a person? And we had some decent technical ideas. But I always remember at the back of the room one day, you have this line of people against the wall, and it was AOL, a person from AOL, Microsoft, and Yahoo.

And people who are geeks on this may know about Amy. EarthLink joined. A...

You know, those four. And gave us anti-spam as, as we've sort of seen it to this day, which is, you know, as I look at this current story as, yeah, maybe. Maybe the same sort of thing.

Mm-hmm. But I'll tell you frankly, that's not what we were trying to do, you know? Not to get some coalition of big players to agree to handle the issue on our behalf a certain way, but actually fix the problem.

So this one I see as an indication that similar sort of thing. You know, the, the, these huge platform companies have realized that it's an environmental problem they can't solve it themselves. You know, can they, you know, can they Amy their way out of it this time?

I don't think so. Um, Kate, you said in the earlier block, I think it's a distributed world, and if we can't get trust locally, then no, the big platforms can't fix it either. Yeah, you know, I, like Alan, I'm pretty d- I think Ira, too bad Ira doesn't do the show much anymore, because he was always dubious of alliances.

But in this case, I think this one has legs because there's a lot of economic incentive. I mean, self-economic ex- incen- incentive. So just the, the, the number that jumped out to me, and this, again, this was or- originally reported by Axios, so it was fed to them by, all the principals, and a couple of the companies actually talked to them, including Amazon and I believe Google.

Um, FBI data is basically $16 billion lost to scammers and cyber criminals in 2024, and the number's gonna exponentially increase because of generative AI. So that got their, that got the big tech companies' attention, so that's why they're doing this, and I actually think they are somewh- semi-serious about this. And it's a good comparison that Chris made back to spam.

I think it's, again, these alliances only work when the bottom line is the top priority like it, it is here. A- and to be clear, you know, and I'm picking on the Amy folks or, or these folks. You know, the individuals are, are doing their jobs.

The companies are where they are. I, and, and- Yeah ... they probably should, but again, I don't think this is something this time that can be solved by the big three or six or 10.

You know what I smelled a little of here when I dug into this? " Yeah. Self-regulation.

Right? If we self-police, if we self-police, we won't have to worry about some outside authority slapping our hands or telling us what we can and can't do. Like the old explicit lyrics on, on records or Yeah, the explicit lyrics label.

So the, so the, you make... That's, that's a really good point, because Meta, Microsoft, Google, Amazon, the companies that were being grilled by congressional leaders would always come up with the answer, "Well, there is no national legislation, but if you trust us, we will self-regulate it. Just trust us.

" And, in a sense, what they're do- trying to do is get ahead of this. And it's again, you're right, Alan, it is the self-regulation trick. I agree.

I agree. Um, but l- let us back up for a second, though. Make no mistake, the, the AI scam- the AI-powered scams are...

I, I got a phish letter today addressed to finance. " And below it was a, a full-blown scammed up copy of an email purporting to be from me to them saying, "Yes, I agree. This is a valid debt.

"That was me, Alan. With the whole email chain of, of- I sent that one, Alan. I sent that.

That was you, Mitch? Oh. That was me.

All right, well that explains it then. But, you know, this was, this was... And so it had like a whole email trail of f- like four or five emails going back and forth that when you looked in the history- Wow ...

you saw. This, this was, this was good. I mean, this, you know, this was a level above the usual nonsense you see from Nigeria or something.

Don't you think we could ask- I had to translate- ... AI to create that for us right now? You could say, "I want it, here's what I want," and create, and maybe the model wouldn't do it, but I don't know.

I think that's- I, I wanna tr- translate a little cybersecurity, you know, community stuff for the rest of you. You know, when Kate and I are smiling over things like this, people like us, it, there's a popcorn emoji in our heads, right? " And on the cybersecurity side, it's like, look, you know, as, as has been discussed on this show week to week, if someone has the intent, and Mitch, to your point, they're doing DevSec properly, and the attackers tend to- ...

having a lot of trust in their coding, they just need to look at you. They just have the intent of breaching you. They'll find all your vulnerabilities and get through.

Uh, phishing, absolutely. You know, the obs, security through obscurity to the frustration of Kate and I and our, and our peers, has actually been pretty effective all these years. Uh, not anymore.

If you are not implementing good security, if you don't have trust built into your systems that's actually operational in the way, again, adult humans actually manage to do in our lives with all the complexity, yeah, bad actors will just push their intent through our- ... our terrible implementations and make us, you know, come up with better solutions. Because when you come up with better solutions, your costs go down.

You're not s- you know, falling for the scams. Mm. You're not spending money on them.

You get to focus on your business, and however you did it becomes the evolutionary force. What's funny, Chris, is, when you were talking about sitting back and, you know, eating popcorn, I was like actually thinking of, you know, 'cause AI didn't invent scams. It only industrialized it.

And what I thought to myself is, you know, it would be like AI has to fight AI scams, and this is like a rock 'em, sock 'em robot type of- ... you know, thing happening. And, you know, I think the only one to sh- to take down the AI scams is AI, you know?

It's- Yeah. There you go. Yeah.

You know? No, well, well, you know, but that is, that is gonna be a security axiom going forward, right? You're gonna need AI to beat AI, to beat AI.

Yeah. Yeah. I am totally- You do But you need AI to secure your systems, right?

There's, you know, a- and there's too much information, and Mitch, you talked about it in DevSec, but you can't read through and put together and correlate all of your network diagrams and configurations, and every poss- you can't. Too much. AI can.

Vector stores can. You can get that coherence to s- to, to implement something really good and fix a lot of these problems, and then AI for defense, I- we'll see. But fixing your problems with AI is a thing, absolutely.

Agreed. Yeah. Well, guys, if that's it, I, I, we could wrap up this version of the gang.

Uh, of course Mike Brazar came down from the high castle to go somewhere today, so he's not around. He went to Boston, didn't he? Oh, I know.

I'm not sure. Did he go to see a Boston- But I'm not sure if he's here today or tomorrow, but we, we, we, we thank him for suggesting the topics, though, so I wanted to acknowledge that. Um, Chris, Kate, John, Mitch, thanks for joining.

Thanks for the, the discussion. Thank you for watching. We've got our Techstrong TV replay, going on now.

Well, actually, it's, it's probably playing already. Uh, we've got a lot going on here. If you get a chance, do subscribe to this new Techstrong newsletter brief.

I think we're gonna put it out on LinkedIn too, or at least a shortened version of it on LinkedIn, so that ought to be fun. Um, we'll be back tomorrow with more gang though, more interesting stuff, probably a lot more AI too. Uh, until then though, this is Alan Schimmel.

We're out.